Image Chroma Enhancement with Quality Preservation

Home , YCbCr

International Journal of Engineering Research & Technology (IJERT) ISSN: 2278-0181 Vol. 3 Issue 8, August - 2014 Image Chroma Enhancement with Quality Preservation

Haripriya A P Riyas K K MTech student (Communication Engineering Asst. Professor, Department of ECE, and signal processing), Department of ECE, Govt. Engineering College Thrissur, Govt. Engineering College Thrissur, Thrissur District, Thrissur District, Kerala, India Kerala, India

Abstract— The goal of Image enhancement in general is to automated image processing techniques. Image enhancement provide a more appealing image or video to achieve more vivid processes consist of a collection of techniques that seek to or realistic colors and to increase the visibility of details. The improve the visual appearance of an image or to convert the chroma enhancements are intimately related to different image to a form better suited for analysis by a human or a attributes of visual sensation and it helps to achieve the best machine [1]. possible combinations of colorfulness and contrast in an efficient The goal of chroma enhancement can be either to increase manner. In practice one of the key issues in the image chroma the colorfulness, or to increase the saturation. The best way to enhancement applications is the strong contrast reduction of the increase the chroma without a lightness change is to use a enhanced image. While there exists several contrast more sophisticated color model, such as CIELAB or enhancement approaches for improving the image contrast, it is CIECAM02, rather than YCbCr. Any color algorithms using still an active field of research. Nowadays most of the High Definition industrial cameras and projectors offer images with the CIELAB or CIECAM02 models guarantee the same color high chroma details. Hence chroma enhanced images along with quality regardless of display color characteristics because the high visual quality is an important challenge in industrial image color characteristics of the device are considered during color processing applications. Two problems are addressed in this value calculations. However, such color models are too paper in order to achieve a better chroma enhanced image. complicated to be implemented in television applications. Firstly, how to preserve the contrast of the image while Hence the color space YCbCr is used for this study. enhancing the image chroma details and secondly, how to The YCbCr color space [2] is one of the most commonly improve the overall image quality details inorder to produce more aesthetically pleasing result. This paper introduces a newIJERTIJERT used color spaces for video coding or image processing. The contrast preserved chroma enhancement algorithm using Y signal is known as “luma” and Cb and Cr are known as YCbCr color space whish solves the problem of contrast color difference or chroma signals. It is not an absolute color reduction while enhancing the image chroma. An adaptive local space; rather, it is a way of encoding RGB information. tone mapping algorithm is also proposed, which further During the chroma enhancement process of an image using improves the overall image quality. To evaluate the YCbCr, each pixel’s Cb and Cr values are increased by some performance, these algorithms are applied on a set of test images constant values. Fig. 1 shows an example in which the left and few evaluation parameters are calculated, which shows that image is the original image and the right is the manipulated the proposed algorithm can effectively enhance the perceived image such that each pixel’s Cb and Cr values are increased image chrominance with image quality preservation compared to other conventional techniques. 1.2 times. Compared to the original, the chroma enhanced image appears more colorful, but also brighter, even though Keywords—image chrominance, contrast enhancement, chroma no lightness change was intended. Such an unintended enhancement, tone mapping, YCbCr color space lightness change also affects image contrast. When the lightness level increases, the image contrast automatically I. INTRODUCTION decreases. This is a serious issue in most of the chroma enhancement cases. “A picture is worth a thousand words” refers to the notion that a complex idea can be conveyed with just a single still image. It also aptly characterizes one of the main goals of visualization, namely making it possible to absorb large amounts of data quickly. Image processing is an important component of modern technologies because , human depends so much on the visual information. About 99% information we perceive about the world is obtained with our eyes. In other words, images are better than any other information (a) (b) form for us to perceive. The enhancement of images helps to improve the interpretability or perception of information in Fig.1. The Effect of Chroma Enhancement Using the YCbCr Color Space images for human viewers or to provide better input for other ( (a) Test Image (b) Chroma Enhanced Image) )

IJERTV3IS081024 www.ijert.org 1515

(This work is licensed under a Creative Commons Attribution 4.0 International License.) International Journal of Engineering Research & Technology (IJERT) ISSN: 2278-0181 Vol. 3 Issue 8, August - 2014

Therefore a contrast preserved chroma enhancement approach Yang et al proposed a hue-preserving graphical algorithms using YCbCr signals that preserve the original approach involving scaling and shifting, that bypassed image contrast while enhancing perceived chroma are computationally intensive coordinate transformation during proposed in this study. Also the tone mapping technique is color image processing [8]. This method was intended for performed on the chroma enhanced contrast preserved cases where only the luminance or only the saturation images. The main goal of the tone mapping technique is to component needed to be modified. provide aesthetically pleasing images and maximizing the In 2004 Meylan et al proposed a retinex-based method for image contrast. Hence the overall image quality can be high dynamic range rendering of natural images [9]. Here the improved along with the enhancement of chroma value. global tone mapping was applied on the linear image and This paper is organized as follows. Section II describes the Luminance component was computed. An adaptive filter literature review undergone on the previous works done in algorithm based on retinex was applied to luminance data. The these fields. Section III discusses one of the previous work in method reduced artifacts like black halos around light sources, this area for the purpose of comparison. Section IV describes but the processing time increased significantly. Most of the the proposed method, chrominance enhancement with contrast work on color reproduction in tone mapping is focused on preservation using YCbCr color space. Section V describes preserving color appearance of the real-world scene, as it is the tone mapping method. Section VI gives the Experimental perceived by the human eye, on a computer screen. The and simulation results using MATLAB. Section VIIconcludes common approach to color treatment in tone mapping, the work. introduced by Schlick [10] , is preserving color ratios. Later papers on tone mapping, employing stronger contrast II. LITERATURE REVIEW compression, observed that the resulting images are over- saturated. The trilateral filter-based HDR tone mapping Most of the published works to date focus on color technique proposed by Choudhury et al is one of the best enhancement in digital color images. Many of these developed techniques for compressing the dynamic range techniques can theoretically be implemented for video as well. [11]. This method can preserve edges and smooth high- However, hardware implementation issues can impose serious gradient regions. However, the main limitation of this method restrictions for many of these techniques. is its high computational cost. In one of the earliest papers on color enhancement, Sachin P. Kamat proposed a Low Bandwidth YCbCr Data Strickland et al [3] recognized that RGB color space did not processing technique for video applications in hand held correspond with the human color perception and so, Devices. YCbCr optimization technique is used in this enhancement algorithms applied directly to RGB images algorithm for additional frame level optimizations. The could lead to color artifacts. Enhancing luminance alone could method significantly reduces the overheads involved in data lead to color artifacts in low luminance regions, and thus transfer and computations thereby improving the performance simultaneous saturation processing was required for proper and thus reducing the power consumed. This paper shows that enhancement. Because of the nonlinear transformation the rough correspondence between the YCbCr signals and between the two color spaces, some processed colors wereIJERT atIJERT visual attributes makes the YCbCr color space more risk of being out-of-gamut when converted back to RGB. In convenient for image processing than other color spaces. 1994, Hague et al [4] presented an approach where histogram equalization was performed on saturation for each hue in the A great deal of research has been carried out to understand image, taking into account the maximum allowable saturation colour appearance phenomena and to model colour for a given luminance. This paper shows that Histogram appearance. In 1997, the CIE recommended a colour equalization on the intensity component can improve the appearance model designated CIECAM97s. In 2002, a new contrast, but can cause desaturation in areas where histogram model CIECAM02 [12] was recommended, which is simpler equalization results in a large reduction in intensity. and has a better accuracy than CIECAM97s. The CIECAM02 was developed to predict color appearance under different In 2001 Lucchese et al [5] presented a two-stage method viewing conditions, especially one for the computer display. for color contrast enhancement based on xy chromaticity The most sophisticated model is CIECAM02 ratified by the diagram. All colors with positive chroma values were CIE (International Commission on Illumination) which maximally saturated through shifting to the borders of a given predicts not only the perceived brightness, but also chroma, color gamut. In the next stage, the colors were desaturated and hue under several illumination conditions. towards a new white point by an appropriate color-mixing rule. In another attempt to develop an image enhancement method based on the chromaticity diagram, Colantoni et al [6] III. CONTRAST ENHANCEMENT IN THE used λSY color space for colorfulness enhancement. This COMPRESSED DOMAIN space is derived from xyY color space and is based on the Contrast enhancement produces an image that subjectively dominant wavelength λ, saturation S and intensity Y. An looks better than the original image by changing the pixel important limitation of the method is a potential for hue shift intensities. This method is a compressed domain image because of the curvilinear nature of the constant perceived hue enhancement technique which works with the DCT lines in the chromaticity diagram. coefficients. The DCT coefficients are separated into various Samadani et al [7] proposed a method for lightening or bands and enhance them using a scaling factor. It uses DC darkening of an image where colors were directly adjusted by coefficients of the Y component for adjust the background moving them along specific lightness-saturation curves while illumination. To preserve the Local Contrast, the AC leaving the hue unchanged. But this method involves only coefficients of the Y component are scaled appropriately. Let lightness adjustment and no color enhancement. In a different

IJERTV3IS081024 www.ijert.org 1516

the scale factor for the DC coefficient be k d and the scale factor for the AC coefficients is for a DCT block Y. The ka processed DCT block is given by (1), Ye

Ye (i , j) = kd Y (i, j), i =j =0 (1)

k a Y (i , j), otherwise

The contrast of the processed image then becomes a / d times of the contrast of the original image. In this algorithm d = a = k for preservation of the contrast. Similarly for the preservation of color vectors in the DCT domain, the

luminance component Y of an image is uniformly scaled by a factor k. Let U and V be the DCT coefficients of the Cb and Fig.2. Block diagram of the Proposed Contrast-preserved Chroma enhancement method Cr components, respectively. If the luminance component Y of an image is uniformly scaled by a factor k , the colors of the processed image with Ye , Ue and Ve are preserved by the B. Analysis of the YCbCr in CIECAM02 following operations: The YCbCr color space is analyzed using CIECAM02 color appearance model in order to develop the algorithm. U (i, j) Ue (i, j) = N (k ( - 128)) + 128 , i=j=0 (2) The two major parts of ciecam02 are its chromatic adaptation N transform, CIECAT02 and its equations for calculating k U(i , j) , otherwise mathematical correlates for the six technically defined dimensions of color appearance: brightness (luminance), V (i, j) lightness, colorfulness, chroma, saturation, and hue. In order V (i, j) = N (k ( - 128)) + 128 , i=j=0 (3) e N to analyze the YCbCr color space in a uniform color space, it is set such that Y has a value between 0 and 255 and Cb and k V(i , j) , otherwise Cr have values between – 128 and 128. The chroma and hue angle in the YCbCr color space are defined as,

2 2 IV PROPOSED CONTRAST PRESERVED Y CC _ chroma = √ c b + cr (4) CHROMA ENHANCEMENT METHOD Y CC _ hue angle = arc tan (Cr/Cb) (5) A contrast preserved chroma enhancement algorithm using YCbCr signals is proposed in this paper, by predicting the amount of luma change needed to compensate for IJERT theIJERT Each YCbCr value is then converted to the lightness change induced by chroma enhancement. Firstly, the CIECAM02 color coordinates J, C, and h values respectively. YCbCr color space is analyzed using a CIECAM02 color For this calculation, the RGB-to-YCbCr conversion is appearance model in order to develop the algorithm. performed. Then YCbCr is analyzed in the CIECAM02 color CIECAM02 is the latest standard color appearance model that space. predicts the perceived lightness J, chroma C, and hue angle h, CIECAM02 defines three surroundings - average, dim, and was proposed in 2002 by CIE. The luma change amounts and dark, which is given in Table I. A surround is a field needed to compensate for the lightness change by the chroma outside the background and outside the white border enhancement are calculated and then modeled. A contrast- (reference white). Surround includes the entire room or the preserved chroma enhancement algorithm is developed using environment. Surround is not measured directly; rather the the lightness change prediction model and its performance is surround ratio is determined and used to assign a surround.

evaluated. TABLE I. A. Block Diagram PARAMETER DECISION TABLE OF CIECAM02

Fig.2 shows the block diagram of the proposed contrast Surround Surround F c Application N preserved chroma enhancement algorithm. The proposed Condition Ratio C algorithm consists of two main functions: chroma Average 1.0 0.69 1.0 Viewing S >0.2 enhancement function and lightness compensation function. R surface colors The chroma enhancement function enhances the input image chroma value, by multiplying it with a chroma weight w. This Dim 0.9 0.59 0.95 Viewing 0 < S < 0.2 makes the resultant image more colorful and also brighter, R television even though no lightness change was intended. Such an unintended lightness change also affects image contrast. The Dark 0.8 0.53 0.8 Using projector S = 0 lightness compensation function calculates Δ luma value and R in a dark room to be added to the input Y value to yield the same CIECAM02 J value with input color, even after chroma enhancement.

IJERTV3IS081024 www.ijert.org 1517

F is the factor determining degree of adaptation, c is the Where L A is the luminance of the adapting field. Using this impact of surrounding and N c is the chromatic induction equation the luminance level adaptation factor F L can be factor. S R is the ratio of the absolute luminance of the calculated. The adapted cone responses are, reference white measured in the surround field to the display 0.42 area. For intermediate conditions, these values can be linearly 400(F R'/100) = L (10) interpolated. R'a 0.42  0.1 (FL R'/100)  27.13

0.42 400(FLG'/100) G'a = 0.42  0.1 (11) (FLG'/100)  27.13

400(F B'/100)0.42 = L (12) B'a 0.42  0.1 (FL B'/100)  27.13

D. Lightness Change Calculation The lightness change induced by the chroma enhancement can be compensated, if the luma value is also changed to a Fig.3. N c and F Variation with c point that has the same lightness as the original color. Therefore, a mathematical model predicting the luma change C. Chromatic Adaptation Transforms amount required to compensate for the lightness change is The most important function of a colour appearance model developed in this work. To develop the model, the luma value is chromatic adaptation transform. CAT02 is the chromatic amount, ΔLuma, which should be added to the original luma adaptation transformation of CIEAM02 model [13]. A CAT to preserve the original lightness, is first calculated. The is calculation procedure is as follows: capable of predicting corresponding colours, which are 1. The input test image which is in RGB format is first defined as pairs of colours that look alike when one is viewed converted into YCbCr color space. under one illuminant and the other is under a different 2. YCC chroma is changed by the multiplication by a illuminant. The CAT02 matrix is given by, given weight, cw. The resulting value will be (Y, cwCb, IJERTIJERTcwCr). 0.7328 0.4296 - 0.1624  3. CIECAM02 lightness J values of the (Y, Cb, Cr) and   M = - 0.7036 1.6975 0.0061 (6) (Y, cwCb, cwCr) data points are calculated.  CAT 02   0.0030 0.0136 0.9834  4. Finally, a ΔLuma value that minimizes the CIECAM02 J difference between the (Y+Δluma, cwCb, cwCr) and (Y, Cb, Inorder to apply the post adaptation transform, the adapted Cr) data points is found. The correlate of lightness J is, RGB responses must be converted from the A M CAT 02 specification to Hunt Pointer Estevez form (HPE). J = 100 cz (13) A w

0. 38971 0.68898 0.07868    M = 0.22981 1.18340 0.04641 (7)  HPE    The achromatic response A is, 0.00000 0.00000 1.0000  A = [2 R'a + G'a + (1/20) B'a - 0.305] Nbb (14) A number of viewing condition components are computed as intermediate values required for further computations. z = 1.48 + √n (15) These include the following factors.

Yb k = 1/ (5L A + 1) (8) n = (16) Yw F = 0.2 K 4 (5L ) + 0.1( 1- K 4 ) 2 ( 5L )1/3 (9) L A A The value n is a function of the luminance factor of the background and provides a very limited model of spatial color appearance. The value of n ranges from 0 for a background luminance factor of zero to 1 for a background luminance

IJERTV3IS081024 www.ijert.org 1518

factor equal to the luminance factor of the adopted white parameters. In other words, the effect of the algorithm point. changes in each pixel according to the local features of the image. Those algorithms are more complicated than the V TONE MAPPING METHOD FOR VISUAL QUALITY global ones; they can show artifacts (e.g. halo effect and ringing); and the output can look unrealistic, but they can ENHANCEMENT provide the best performance, since human vision is mainly Tone mapping is a major component of image sensitive to local contrast. reproduction. It provides the mapping between the light emitted by the original scene and display values. Tone B. Tone mapping of Intensity Component mapping is a technique used in image processing and In order to preserve the original colors, the tone mapping is computer graphics to map one set of colors to another in applied only to the intensity component and the output color order to approximate the appearance of high dynamic range image is obtained by nonlinear transformation of the original images in a medium that has a more limited dynamic range. color components. The intensity component of the input Print-outs, CRT or LCD monitors, and projectors all have a image is denoted as Y (i, j) and the output pixel intensity is limited dynamic range that is inadequate to reproduce the full denoted by Y (i, j), with (i, j) being the pixel coordinates. range of light intensities present in natural scenes. The 0 Human Visual System (HVS ) processes the scene radiances The following function is utilized to map the values of Y (i, j) in a non-linear manner through different adaptation to the new values Y 0 (i , j)[15] , processes. Electronic devices capture the scene radiances linearly. A tone mapping operator is necessary to non-linearly max Y  R encode the image as well as to map it to the display m(i. j) Y 0 (i, j) = Y(i, j) (17) characteristics so that the displayed image corresponds to our Y(i, j) Ym (i, j)  R memory of the original scene. Where max is the maximum value of Y (i, j), Y m (i; j) is the smoothed version of Y (i , j) and R is a regularization term that controls the global effect of the mapping process. The parameter R is selected as half of the average of Y (i , j) . The reason for this selection is that small values of R will strengthen the global tone mapping effect and increase more the brightness of the processed image compared to a larger R.

B. Tone mapping Regularization In order to introduce proposed method lets first analyze Fig. 4 . Processing of Scene radiances by HVS and Tone mappingIJERT (17), where the function used to transform the intensity of the operator IJERTinput image is defined. The effect of (17) for two different values of the regularization R is computed first. Let us take A. Purpose and Methods of Tone mapping Y (1) be the intensity computed using R1 and Y (2) be the The goals of tone mapping can be differently stated 0 0 depending on the particular application. In some cases intensity computed using R2: producing just aesthetically pleasing images is the main goal, while other applications might emphasize reproducing as (1) maxYm(i. j)  R1 many image details as possible, or maximizing the image Y 0 (i , j) = Y(i, j) (18) contrast. The goal in realistic rendering applications might be Y(i, j) Ym (i, j)  R1 to obtain a perceptual match between a real scene and a (2) maxYm(i. j)  R2 displayed image even though the display device is not able to Y 0 (i , j) = Y(i, j) (19) reproduce the full range of luminance values. Various tone Y(i, j) Ym (i, j)  R2 mapping operators have been developed in the recent years. C. Proposed Tone mapping algorithm They all can be divided in two main types - Global and Local methods [14]. In order to implement adaptive tone mapping method, the The global tone mapping algorithms have low following procedure is implemented: computational complexity but provide lower visual quality  The input color image with the red, green, and the blue compared to the local tone mapping solutions. In the case of components are transformed first in the YUV domain global tone mapping solutions usually the histogram of the and only the Y (i, j) component is processed. pixel values are build and a tone mapping function is  Estimate the optimum size for each pixel Y (i, j). A set of obtained based on the histogram. The local tone mapping M different window sizes are selected. For a given pixel methods usually ensure a better visual quality of the Y (i, j) compute the corresponding averages Y (i, j) for processed image but at the expense of increased m computational complexity and processing power. The each of the pre-defined window sizes N k ,k = 1, . . . ,M . parameters of the non-linear function change in each pixel,  The image of the local averages Y (i , j) is computed according to features extracted from the surrounding m using the optimum local window sizes.

IJERTV3IS081024 www.ijert.org 1519

 The computed local averages Y (i , j) are utilized to Fig. 5. Chroma weight distribution (a) blood cell test image , (b) chroma m enhanced imagen with w = 0.8 , (c) w = 0.9 , (d) w = 1.0 , (e) w = 1.1 , estimate the optimum value. (f) w = 1.2 , (g) w = 1.3 , (h) w = 1.  The tone mapped component Y (i , j) is computed B. Comparison with Conventional Method 0 using Y m (i , j) and optimum value R opt The chroma enhanced images are also analyzed in an existing compressed domain contrast enhancement method. The comparison of the proposed method with this max Ym(i. j)  Ropt Y 0 (i , j) = Y(i, j) (20) conventional method is also performed in order to show that, Y(i, j)  Ym (i, j)  Ropt in general, the proposed algorithm outperforms the conventional chroma enhancement technique.

VI EXPERIMENTAL RESULTS

This section deals with the portrayal of the simulation results obtained from the mat lab during the development of proposed method, chroma enhancement with image quality preservation. The proposed algorithm is tested on a set of images of different compositions downloaded from the standard image databases. It is found that the proposed algorithm gives best result compared to other conventional enhancement methods. Image chroma enhancement with contrast preservation is performed first and the performance is evaluated. Thereafter Tone mapping algorithm for improve the visibility details of the contrast preserved image is also developed using mat lab and the performances are evaluated.

A. Test Image preparation

Four different images shown in given Fig.7 were selected as test images. The resolutions of those images are set as Fig. 6. Comparisons of the Chroma enhanced contrast preserved images 500X500 pixels respectively. Each images chroma was then using the conventional and proposed algorithm increased by some weight. The chroma weight can be changed from 0.7 to 1.3. During the chroma enhancement C. subjective evaluation of proposed method process of test image using YCbCr, each pixels Cb andIJERT CIJERTr values are increased by some constant value chroma weight The images in given figure clearly shows the effect of the (w). The chroma weight can be selected as any constant proposed algorithm. The original test images are in Fig.7(a). values. The best possible chroma weights are varied from 0.5 The test images chroma values are enhanced by a weight of (50%) to 1.3 (130%) with a 0.1 increment. The effective 1.2 for each pixels. As a result image contrast decreases due chroma variation occurs between 0.7 to 1.3. In this work a to an unintended lightness change. These results are shown in chroma weight of 1.2 is selected for the effective chroma Fig.7 (b) . The results obtained using the proposed method enhancement. The image color appearance is changed along are shown in Fig.7 (c). The chroma enhanced images are also with the chroma variation. The chroma variation effects in an analyzed in an existing compressed domain contrast image are shown in Fig.5. enhancement method. The simulation results are shown in Fig.6 . The chroma enhanced images using the conventional compressed domain method look notably lighter and have lower contrast compared to the proposed method. The images resulting from the proposed algorithm revealed that, in general, the proposed method outperforms the conventional contrast enhancement technique, especially for images containing red colors or fine textures.

D. Numerical algorithm Performance test The goal of the proposed algorithm is preserving the original lightness while the chroma value increases. As a measure, PSNR (peak signal-to-noise ratio) values were calculated for the results in both conventional and proposed method for the different test images. Table II lists the average PSNR values for all four test images. It is clear that there is an image dependency, even though the PSNR values of the

IJERTV3IS081024 www.ijert.org 1520

proposed algorithm images are generally higher than those of VII CONCLUSION the conventional algorithm. A chroma enhancement algorithm with the preservation of original image characteristics is proposed in this paper. The effect of chroma enhancement on image contrast is considered first. The amounts of luma change needed to compensate for the lightness change by the chroma enhancement are calculated and modeled. This research suggests the possibility that color enhancement algorithms could be developed using the YCbCr color space with performances that are equivalent to the performance achieved when a uniform color space is used ,such as in CIECAM02 . Numerical tests are conducted by calculating PSNR of the resultant images. The results revealed that, in general, the proposed algorithm outperforms the conventional chroma enhancement techniques, especially for images containing red colors or fine textures. In the next section a novel method for local tone mapping of color images is proposed. This method uses two adaptive control parameters to increase the visual quality of the processed images. One parameter is the spatial range of the local adaptation, that is the optimal size N of the neighborhood affecting the global tone mapping effect. The other parameter, R, controls the strength of the local tone mapping function. This value is optimized based on the Fig.7. Proposed Contrast preserved Chroma enhancement algorithm image variance. Performance comparison of all the methods Results (a) Test images (b) Chroma enhanced images (c) Image using based on Peak Signal to Noise Reduction is to be carried out, proposed method (d) Image using Tone mapping method which reveals the efficiency of the proposed method. Thus chroma enhancement of images along with the preservation It is clear that the PSNR values of the proposed algorithm of contrast with better visibility details can be achieved with images are generally higher than those of the conventional the proposed method and, which can solve practical contrast algorithm, which shows the efficiency of the proposed reduction problem due to chroma enhancement effectively. method. This thesis considers the lightness changes induced by the YCC chroma change. However, other color distortions also 2 IJERToccur due to chroma enhancement. The changes in the luma PSNR = 10 log ( 255 / MSE ) (21)IJERT value produce unwanted chroma and hue changes in addition to the desired lightness changes. Therefore, further research is Where MSE represents the mean squared error . The average required to determine a method that can compensate for all PSNR values for all four test images are given below. such unwanted color distortions by using the YCbCr color space. Also the local tone mapping algorithm involves TABLE II. increased processing power and computational complexity. PSNR (dB) OF THE CONVENTIONAL AND PROPOSED METHODS Therefore local tone mapping algorithm has to be improved further to reduce the computational complexity PERFORMANCE TEST – PSNR(dB)

Test Conventional Proposed Tone Images Compressed ciecam02 mapping VIII REFERENCES domain method Method [1] T. Leisti, J. Radun, T. Virtanen, R. Halonen, and G. Nyman, “Subjective (a) 11.83 24.67 28.47 experience of image quality: attributes, definitions, and decision making of subjective image quality,” IS &T/SPIE Electronic Imaging, pp. 72420 –72420, January 2009. (b) 13.26 22.64 25.87 [2] S. P. kamat, “Low bandwidth YCbCr data processing technique for video applications in handheld devices,” IEEE Transactions on Consumer Electronics, vol. 56, no. 3, pp. 1770 – 1774, August 2010. (c ) 11.46 16.63 18.27 [3] R. N. Strickland, C.-S. Kim, and W. F. McDonnell, “Digital color image enhancement based on the saturation component,” Optical Engineering, (d) 15.86 26.24 28.32 vol. 26, pp. 609–616, July 1987. [4] G. E. Hague, A. R. Weeks, and H. R. Myler, “Histogram equalization of the saturation component for true-color images using the CY color space,” Proc. SPIE 2298, Applications of Digital Image Processing XVII, vol. 2298, September 1994. [5] Lucchese, Mitra, and J. Mukherjee, “A new algorithm based on saturation and desaturation in the XY chromaticity diagram for enhancement and

IJERTV3IS081024 www.ijert.org 1521

re-rendition of color images,” International Conference on Image Processing, vol. 2, pp. 1077 – 1080, October 2001. [6] P. Colantoni, N. Bost, and A. Tremeau, “Colorfulness enhancement in λsy color space,” Second European Conference on Color in Graphics, Imaging, and Vision and Sixth International Symposium on Multispectral Color Science, vol. 14, pp. 161–166, January 2004 [7] R. Samadani and G. Li, “Geometrical methods for lightness adjustment in YCC color spaces,” Proceedings of SPIE, pp. 15–19, January 2006. [8] C. C. Yang and J. J. Rodriguez, “Saturation clipping in the color spaces,” Proceedings of SPIE 2658, Color Imaging: Device-Independent Color, Color Hard Copy, and Graphic Arts, vol. 2658, pp. 297–307, March 1996 [9] L. Meylan and S. Susstrunk, “High dynamic range image rendering with a retinex-based adaptive filter,” IEEE Transactions on Image Processing, vol. 15, no. 9, pp. 2820–2830,September 2006. [10] C. Schlick, “Quantization techniques for the visualization of high dynamic range pictures,” In Photorealistic Rendering Techniques, Eurographics,Springer -Verlag Berlin Heidelberg NewYork,pp.7– 20,1994. [11] P. Choudhury and J. Tumblin, “The trilateral filter for high contrast images and meshes,” Eurographics Symposium on Rendering, pp. 1–11, June 2003. [12] C. J. Li and M. R. L. andR. W. G. Hunt, “A revision of the ciecam97s model,” Color Research and Application, vol. 25, pp. 260–266, June 2000. [13] I. Tastl, M. Bhachech, N. Moroney, and J. Holm, “ICC color management and ciecam02,” Color and Imaging Conference Final Program and Proceedings, pp. 217–223, January2005. [14] K. Smith, G. Krawczyk, K. Myszkowski, and H.-P. Seidel, “Beyond tone mapping: Enhanced depiction of tone mapped HDR images,” In Proceedings of Computer Graphics Forum, pp. 427–438, 2006.

[15] L. Meylan, D. Alleysson, and S. Susstrunk, “Model of retinal local adaptation for the tone mapping of color filter array images,” Journal of the Optical Society of America A (JOSAA), vol. 24, no. 9, p. 28072816, September 2007.

IJERTIJERT

IJERTV3IS081024 www.ijert.org 1522

(This work is licensed under a Creative Commons Attribution 4.0 International License.)